Quick favor: Our records indicate that you aren’t opening this email. But records can be wrong. Please click here if you’d like to remain subscribed to Fintech Takes. 

Happy Wednesday, Fintech Takers!

I hope your week has, so far, exceeded your expectations.

Mine certainly has. We are cranking on content right now. Got lots of good stuff coming your way, including podcasts … like the one that dropped today!

— Alex  

P.S. — Speaking of exceeding expectations, those first two NBA play-in tournament games were absolute barn burners! I’m glad the Hornets are still in. The NBA is more fun with that team still in the hunt (though that uncalled foul by LaMelo Ball wasn’t great). And Portland feels like it will be a legit annoyance to the Spurs. Great stuff all around.  

Was this email forwarded to you?

3 BIG IDEAS FROM THE PODCAST

This week on Not Fintech Investment Advice, Simon Taylor and I are back at it.

We brought four companies to the table. 

We talked about accounts receivable. We talked about wealth management. I offered up some fun lending infrastructure companies that broke Simon’s brain a bit. 

Our conversation was infused with AI, as everything is these days, but it wasn't just about AI.

And read below for my three big ideas... 

Lending insurance isn’t new (if you lose your job or become disabled, there are products that cover your payments). Fraud indemnification isn't new. Chargeback protection has existed for years.

But I had never before encountered a company offering insurance on the credit underwriting model itself.

That's what MKIII (pronounced Mark 3) is doing. Community banks and credit unions embed MKIII’s underwriting model into their digital lending workflow. The model approves borrowers that those institutions might’ve otherwise declined. And if the model drifts or underperforms in a way that results in losses, the embedded insurance covers them.

Simon's instinct was to ask what the model is backtested on, which is the right question. Lending is a learning business, as I like to say around here, and MKIII’s unique approach would certainly require a lot of learning to work well. The company claims that it has trained its model on $3 billion-worth of loan data, representing 20+ years of performance history (which, I’m guessing, means that the training data was supplied by one of the credit bureaus, but that’s just a guess on my part).

The more interesting implication, to me, is what this does to the balance sheet constraints that hold smaller lenders back. 

Banks and credit unions are required to hold reserves against expected losses, capital set aside against the loans that may go bad. If you can insure the model's performance, you potentially unlock relief on those reserve requirements. Lending looks a little less like a balance sheet business and more like a transactional business. 

In an era where private credit is growing and the lending value chain is increasingly modular, MKIII certainly appears to be skating to where the puck is going.

I’ve not yet seen an agentic AI product in financial services that genuinely puts its users’ needs above everything else. A true “guardian agent,” as some call it. Something that isn't beholden to advertisers or corporate parents. Something that charges you directly for its objectivity, and that's focused on your long-term ambitions rather than your next transaction. 

All the big PFMs that have gotten traction over the last decade converged on paid subscriptions as the business model. They all learned from Mint. The subscription model is the only one that keeps the PFM providers’ incentives honest.

I think this same basic reality will need to apply to the next generation of agentic AI-focused PFM solutions as well, which is why the AI financial planner Wealth Architect is interesting. 

Wealth Architect does several things well, including but not limited to natural language goal-setting, tax-aware scenario planning, Monte Carlo stress testing, and transaction categorization. The interface is exactly what financial planning should look like: someone you can ramble at in whatever half-formed language you use to think about money, and who translates that into something actionable.

The problem is that the competitive environment is now genuinely ferocious. Revolut announced a personal finance agent the day we recorded this episode. Cash App has one. Public.com has one. Plaid recently partnered with Perplexity. OpenAI just bought Hiro. Shoot, you could just instruct Claude to scrape your bank accounts today and assemble a full financial picture, without going through any data aggregator or PFM tool at all.

Wealth Architect is pointing at the right target. The question is whether any standalone tool can establish that center of gravity before larger financial services apps and general-purpose AI chatbots swallow this use case whole.

Traditional credit underwriting models were (for the most part) built for a specific type of worker, one who’s employed full-time, paid on a regular schedule, documented by W-2s and pay stubs. 

A company called CloutScore is betting that this kind of worker is a shrinking fraction of the American economy, and the numbers on their website make a compelling case: 76 million Americans now earn income outside traditional employment, 27 million are full-time independent workers (projected to hit 86 million by 2027).

CloutScore’s bet is that underwriting infrastructure hasn't caught up, and I think they're right. 

The bank accounts of gig workers and content creators tell an incomplete story. Someone earning across Uber, Lyft, Etsy, and YouTube has income fragmented across multiple platforms, paid on different schedules, with no single document that summarizes any of it. CloutScore pulls and distills this data into a 300-to-850 score that lenders can use as an input into the underwriting decisions.

The analogy that felt most accurate to me: this is what Square did with merchant cash advance lending. Rather than using the bank account as a source of truth, Square used the proprietary data it had on its sellers: inventory, sales patterns, customer behavior, etc. That depth of platform-native data produced an underwriting model that traditional SMB lenders were hard-pressed to match.

The open question is how long will platforms like Uber, Shopify, Etsy, and YouTube allow consumer-permissioned access to their data?

This is the same fight that's been playing out in open banking for years, and it's eventually going to happen here too. Whether CloutScore can build enough of a lender network before that moment arrives remains to be seen.

🎬 DIRECTOR'S COMMENTARY

Do you remember Klout, the social media analytics tool that allowed users to rate their online social influence via the "Klout Score", which was a numerical value between 1 and 100?

CloutScore feels, on a superficial level, like Klout’s successor. 

I wonder if the founders of CloutScore thought about that when they picked the name.

WHAT I'M LISTENING TO

Lots of good AI talk in this one.

I’ve been enjoying this entire series on the economics of books and the publishing industry, but this episode in particular resonated with me. I love bookstores, and my wife (who I also love) used to work in one. 

Thanks for the read! Let me know what you thought by replying back to this email. 

— Alex  

Join 2,444 other finance and fintech leaders in the Fintech Takes Network

The conversation doesn't have to stop here
Keep learning and connecting in the Fintech Takes Network

EVENTS  |  FEED  |  LIBRARY  |  DIRECTORY

Get your brand in front of 63,000+ fintech and banking executives.

Workweek Media Inc. 
1023 Springdale Road, STE 9E
Austin, TX 78721

Want to ruin my day? Unsubscribe.